CVE-2026-33626

CVE Enrichment

CVSS: 7.5 HIGH
EPSS: 1.51%
EPSS Percentile: 73.76%
CISA KEV: Not listed

Description

LMDeploy is a toolkit for compressing, deploying, and serving large language models. Versions prior to 0.12.3 have a Server-Side Request Forgery (SSRF) vulnerability in LMDeploy's vision-language module. The `load_image()` function in `lmdeploy/vl/utils.py` fetches arbitrary URLs without validating internal/private IP addresses, allowing attackers to access cloud metadata services, internal networks, and sensitive resources. Version 0.12.3 patches the issue.

Technical Details

CVSS Version: 3.1
CNA / NVD Source: security-advisories@github.com
Published: 21-Apr-2026 04:16
Last Modified: 17-Jun-2026 17:37
EPSS Date: 2026-10-12
CWE: CWE-918

CVSS Vector


            CVSS:3.1                            AV:N /
                            AC:L /
                            PR:N /
                            UI:N /
                            S:U /
                            C:H /
                            I:N /
                            A:N

Affected Products / Versions

internlm / lmdeploy
Version / affected range: * · before (exclusive) 0.12.3
cpe:2.3:a:internlm:lmdeploy:*:*:*:*:*:*:*:*

References

Last enrichment sync: 13-Oct-2026 01:48
Sources: NVD · FIRST EPSS · CISA KEV
Latest successful provider update; providers refresh independently.

Reported Context (1)

CVE that explicitly identifies this as a critical cloud attack vector [10]. Most recently, CVE-2026-33626 in LMDeploy’s AI inference module demonstrated the identical pattern, with in-the-wild exploitation Unauthenticated MCP Server Exposed AWS Credentials Through SSRF and File Inclusion

CVE (5)

Malware (2)

Threat Actors (1)

Vendors (3)

Products (7)

Tools (3)

Industries (4)

Countries (3)

Note: Related entities, including threat actors, malware, CVEs, MITRE ATT&CK techniques, vendors, products, tools, countries, and industries, are shown when they appear in the same reporting. Their presence does not necessarily mean they were targeted, compromised, vulnerable, responsible for the activity, or directly involved in the incident.